• The Journal of Korea Robotics Society
• /
• v.11 no.4
• /
• pp.226-234
• /
• 2016

In this study, a model-referenced underwater navigation algorithm is proposed for high-precise underwater navigation using monocular vision near underwater structures. The main idea of this navigation algorithm is that a 3D model-based pose estimation is combined with the inertial navigation using an extended Kalman filter (EKF). The spatial information obtained from the navigation algorithm is utilized for enabling the underwater robot to navigate near underwater structures whose geometric models are known a priori. For investigating the performance of the proposed approach the model-referenced navigation algorithm was applied to an underwater robot and a set of experiments was carried out in a water tank.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.24 no.12
• /
• pp.1574-1580
• /
• 2020

Traditional pose estimation methods include using special devices or images through image processing. The disadvantage of using a device is that the environment in which the device can be used is limited and costly. The use of cameras and image processing has the advantage of reducing environmental constraints and costs, but the performance is lower. CNN(Convolutional Neural Networks) were studied for pose estimation just using only camera without these disadvantage. Various techniques were proposed to increase cognitive performance. In this paper, the effect of the skip connection on the network was experimented by using various skip connections on the joint recognition of the hand. Experiments have confirmed that the presence of additional skip connections other than the basic skip connections has a better effect on performance, but the network with downward skip connections is the best performance.

• Journal of Broadcast Engineering
• /
• v.24 no.2
• /
• pp.264-272
• /
• 2019

In this paper, we propose a 3D human body reconstruction and refinement method from the frames extracted from a video to obtain natural and smooth motion in temporal domain. Individual frames extracted from the video are fed into convolutional neural network to estimate the location of the joint and the silhouette of the human body. This is done by projecting the parameter-based 3D deformable model to 2D image and by estimating the value of the optimal parameters. If the reconstruction process for each frame is performed independently, temporal consistency of human pose and shape cannot be guaranteed, yielding an inaccurate result. To alleviate this problem, the proposed method analyzes and interpolates the principal component parameters of the 3D morphable model reconstructed from each individual frame. Experimental result shows that the erroneous frames are corrected and refined by utilizing the relation between the previous and the next frames to obtain the improved 3D human reconstruction result.

• Journal of the Korea Academia-Industrial cooperation Society
• /
• v.21 no.7
• /
• pp.285-291
• /
• 2020

In manufacturing, humans are being replaced with robots, but expert skills remain difficult to convert to data, making them difficult to apply to industrial robots. One method is by visual motion recognition, but physical features may be judged differently depending on the image data. This study aimed to improve the accuracy of vision methods for estimating the posture of humans. Three OpenPose vision models were applied: MPII, COCO, and COCO+foot. To identify the effects of face-covering accessories and image preprocessing on the Convolutional Neural Network (CNN) structure, the presence/non-presence of accessories, image size, and filtering were set as the parameters affecting the identification of a human's posture. For each parameter, image data were applied to the three models, and the errors between the actual and predicted values, as well as the percentage correct keypoints (PCK), were calculated. The COCO+foot model showed the lowest sensitivity to all three parameters. A <50% (from 3024×4032 to 1512×2016 pixels) reduction in image size was considered acceptable. Emboss filtering, in combination with MPII, provided the best results (reduced error of <60 pixels).

• Journal of KIISE
• /
• v.41 no.11
• /
• pp.927-934
• /
• 2014

In this paper, we propose an effective method for recognizing daily human activities from a stream of three dimensional body poses, which can be obtained by using Kinect-like RGB-D sensors. The body pose data provided by Kinect SDK or OpenNI may suffer from both the view variance problem and the scale variance problem, since they are represented in the 3D Cartesian coordinate system, the origin of which is located on the center of Kinect. In order to resolve the problem and get the view-invariant and scale-invariant features, we transform the pose data into the spherical coordinate system of which the origin is placed on the center of the subject's hip, and then perform on them the scale normalization using the length of the subject's arm. In order to represent effectively complex internal structures of high-level daily activities, we utilize Hidden state Conditional Random Field (HCRF), which is one of probabilistic graphical models. Through various experiments using two different datasets, KAD-70 and CAD-60, we showed the high performance of our method and the implementation system.

• Korean Journal of Computational Design and Engineering
• /
• v.22 no.1
• /
• pp.89-99
• /
• 2017

In order to make a virtual endoscopy system effective for exploring the interior of the 3D model of a human organ, it is necessary to generate an accurate navigation path located inside the 3D model and to obtain consistent camera position and pose estimation along the path. In this paper, we propose an approach to virtual navigation of blood vessels, which makes proper use of orthogonal contours and skeleton curves. The approach generates the orthogonal contours and the skeleton curves from the 3D mesh model and its voxel model, all of which represent the blood vessels. For a navigation zone specified by two nodes on the skeleton curves, it computes the shortest path between the two nodes, estimates the positions and poses of a virtual camera at the nodes in the navigation zone, and interpolates the positions and poses to make the camera move smoothly along the path. In addition to keyboard and mouse input, intuitive hand gestures determined by the Leap Motion SDK are used as user interface for virtual navigation of the blood vessels. The proposed approach provides easy and accurate means for the user to examine the interior of 3D blood vessels without any collisions between the camera and their surface. With a simple user study, we present illustrative examples of applying the approach to 3D mesh models of various blood vessels in order to show its quality and usefulness.

• Journal of Broadcast Engineering
• /
• v.4 no.2
• /
• pp.176-186
• /
• 1999

In order to produce an image that lends realism to audience in the virtual studio system. it is important to synchronize precisely between foreground objects and background image provided by computer graphics. In this paper, we propose a method of camera parameter extraction for the synchronization by tracking the pose of TV camera. We derive an equation for extracting camera parameters from inverse perspective equations for tracking the pose of the camera and 3-D transformation between base coordinates and estimated coordinates. We show the validity of the proposed method in terms of the accuracy ratio between the parameters computed from the equation and the real parameters that applied to a TV camera.

• Journal of the Institute of Electronics Engineers of Korea CI
• /
• v.46 no.2
• /
• pp.84-91
• /
• 2009

This paper presents an online partial 3D modeling methodology that uses a mobile augmented reality system and aerial photographs, and a tracking methodology that compares the 3D model with a video image. Instead of relying on models which are created in advance, the system generates a 3D model for a real building on the fly by combining frontal and aerial views. A user's initial pose is estimated using an aerial photograph, which is retrieved from a database according to the user's GPS coordinates, and an inertial sensor which measures pitch. We detect edges of the rooftop based on Graph cut, and find edges and a corner of the bottom by minimizing the proposed cost function. To track the user's position and orientation in real-time, feature-based tracking is carried out based on salient points on the edges and the sides of a building the user is keeping in view. We implemented camera pose estimators using both a least squares estimator and an unscented Kalman filter (UKF). We evaluated the speed and accuracy of both approaches, and we demonstrated the usefulness of our computations as important building blocks for an Anywhere Augmentation scenario.

• The KIPS Transactions:PartB
• /
• v.14B no.3 s.113
• /
• pp.163-170
• /
• 2007

In autonomous navigation of a mobile vehicle or a mobile robot, localization calculated from recognizing its environment is most important factor. Generally, we can determine position and pose of a camera equipped mobile vehicle or mobile robot using INS and GPS but, in this case, we must use enough known ground landmark for accurate localization. hi contrast with homography method to calculate position and pose of a camera by only using the relation of two dimensional feature point between two frames, in this paper, we propose a method to calculate the position and the pose of a camera using relation between the location to predict through perspective transform of 3D feature points obtained by overlaying 3D model with previous frame using GPS and INS input and the location of corresponding feature point calculated using KLT tracking method in current frame. For the purpose of the performance evaluation, we use wireless-controlled vehicle mounted CCD camera, GPS and INS, and performed the test to calculate the location and the rotation angle of the camera with the video sequence stream obtained at 15Hz frame rate.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2018.10a
• /
• pp.686-689
• /
• 2018

3D 손 포즈 추정(Hand Pose Estimation, HPE)은 스마트 인간 컴퓨터 인터페이스를 위해서 중요한 기술이다. 이 연구에서는 딥러닝 방법을 기반으로 하여 단일 RGB-Depth 카메라로 촬영한 양손의 3D 손 자세를 실시간으로 인식하는 손 포즈 추정 시스템을 제시한다. 손 포즈 추정 시스템은 4단계로 구성된다. 첫째, Skin Detection 및 Depth cutting 알고리즘을 사용하여 양손을 RGB와 깊이 영상에서 감지하고 추출한다. 둘째, Convolutional Neural Network(CNN) Classifier는 오른손과 왼손을 구별하는데 사용된다. CNN Classifier 는 3개의 convolution layer와 2개의 Fully-Connected Layer로 구성되어 있으며, 추출된 깊이 영상을 입력으로 사용한다. 셋째, 학습된 CNN regressor는 추출된 왼쪽 및 오른쪽 손의 깊이 영상에서 손 관절을 추정하기 위해 다수의 Convolutional Layers, Pooling Layers, Fully Connected Layers로 구성된다. CNN classifier와 regressor는 22,000개 깊이 영상 데이터셋으로 학습된다. 마지막으로, 각 손의 3D 손 자세는 추정된 손 관절 정보로부터 재구성된다. 테스트 결과, CNN classifier는 오른쪽 손과 왼쪽 손을 96.9%의 정확도로 구별할 수 있으며, CNN regressor는 형균 8.48mm의 오차 범위로 3D 손 관절 정보를 추정할 수 있다. 본 연구에서 제안하는 손 포즈 추정 시스템은 가상 현실(virtual reality, VR), 증강 현실(Augmented Reality, AR) 및 융합 현실 (Mixed Reality, MR) 응용 프로그램을 포함한 다양한 응용 분야에서 사용할 수 있다.